Effectively Addressing Control Application

advertisement
Effectively Addressing
Control Applications
Presenters
• Terry Blevins
• James Beall
Typical Single Input-Output Control Loop
Agenda
• Measurements – How to avoid signal aliasing
• The impact measurement status and selected status
options on control.
• Single loop control –selection of structure based on
application requirements, use of PID notch gain option,
providing quick recovery from startup conditions.
• Application of output characterization to provide linear
installed characteristics.
• Feedforward – Use of PID feedforward.
• Constructing a summing or multiplying feedforward
outside the PID block, advantages and disadvantages.
• Implementation and commissioning of dynamic
compensation.
Agenda (Cont)
• Override Control - Implementation and commissioning of
override control using PID’s and control selector
• Cascade control – use of PID dynamic reset option to
improve performance. Implementing MPC cascade.
• Ratio control – Using the Ratio block based on SP or
wild flow. Adjustment of Ratio target, impact of options.
• Split range control and valve position control –
Implementation using standard blocks, impact of options.
• Duty cycle and increase open/close actuation – Use of
the AO with a DO channel to achieve precisely timed
on/off actuation.
• Fuzzy Logic Control for loops that are lag dominant.
• DeltaV Predict MPC block in control applications –
addressing difficult dynamics, impact on single loop and
override control applications.
Processing Analog Inputs
Traditional Transmitter
(4-20ma or Hart)
DeltaV Analog Input Module
Digital Filtered
Value to Controller
A/D
Two-pole hardware filter with
cutoff point (-3db) at 2.7 Hz
Analog-to-digital
Conversion (16 bit)
Configurable
Software Filter
Every 22 milliseconds
I/O Module Software Filter
• Software filtering may be
applied at the I/O module
to avoid aliasing.
• Only required if the 4-20
ma signal contains
frequencies > 1/2X the
control module execution
rate
- Value sampled by the control module
- Aliased signal as see by the control module
- Actual signal (with noise removed)
Filter Guideline
Analog Measurement Limit and Bad
Status
• High or low limit is set in status if the
measurement exceeds the over-range and
under-range values specified for the channel
• The status quality will be set to BAD if the
measurement excees the A/D range (open or
short condition).
Adjusting Over & Under Range Detection
Analog Input Block
Analog Input Block
OUT_SCALE
XD_SCALE
SIMULATE_IN
100 %
100 %
L_TYPE
PV
(Measurement)
FILTER
AUTO
OUT
MANUAL
CONVERT
PV_FTIME
MODE
UNIT
0%
SIMULATE
(VALUE +STATUS)
SQ. ROOT
LOW_CUT
UNIT
0%
MANUAL
VALUE
Status Provided by the Analog Input
• A status quality of Uncertain or BAD may be
created under limit conditions or Man Mode
based on STATUS_OPT parameter selections.
• Status is set to BAD if the block mode is set to
Out-of-Service.
AI Status Options
PID Block
PID Function Block
MODE
BCKCAL_OUT
SP
CAS_IN
OUT_SCALE
C
GAIN, RESET, RATE
A
100 %
IN
0%
BCKCAL_IN
BYPASS
PID
CALC
0%
PV_SCALE
0%
FF_GAIN
FF_SCALE
TRK_IN_D
100 %
TRK_VAL
LIMIT
OUTPUT
+
100 %
FF_VALUE
100 %
0%
TRK_SCALE
Note: For simplicity, not all modes are shown.
OUT_HI_LIM
OUT_LO_LIM
OUT
Impact of IN Status
• Status_OPTS determine if control will continue with
Uncertain status. The recommended default is Use
Uncertain as Good
• If status is BAD, then actual mode will automatically go
to MAN.
• An IN limit status of Constant will automatically cause
the reset action to hold last value (to prevent windup
under this condition).
PID Function Block Algorithm
• Either Series or Standard form may be selected
using the FORM parameter. The default is
Standard. Response is identical of either
selection if rate action is not used.
• Whether proportional and derivative action are
taken on error or PV value may be selected
using the STRUCTURE parameter.
Selection of PID Form
Differences Between Forms
• Standard and Series have identical response for
Proportional and Integral tuning (0 derivative).
• The Series form allows derivative (rate) and
integral (reset) to be changed independently i.e.
non-interacting.
• The Standard Form is capable of a more flexible
response
PID STRUCTURE Parameter
PID STRUCTURE Parameter
• Example of I on error, PD on PV
• Change setpoint without a large disturbance in
the manipulated variable (output)
• Level of a feed tank
• Base level of a column that feeds another
column
PID STRUCTURE Parameter
• Reactor feed tank: PI on error, D on PV
Controller Output – Flow to reactor
PID STRUCTURE Parameter
• Reactor feed tank: I on error, PD on PV
Controller Output – Flow to reactor
Adjustment of Beta & Gamma
• If two degrees of freedom
is selected as the PID
structure, then BETA and
GAMA may be adjusted to
determine the fraction of
proportional and
derivative action on error
vs. PV.
General Block Diagram Of PID
FF
SP
+
Forward
Path
PV
+
M
LIMITER

+ R
RESET
FILTER
Reset created by positive
feedback network.
Automatically provides antireset windup protection
-

L
+
From External
Element
External Reset
• The FRSI_OPTS for “Dynamic Reset Limit” may
be selected to enable external reset; i.e., the use
of BKCAL_IN in the reset calculation.
• When this option is selected, then the
downstream block’s CONTROL_OPTS should
have “Use PV for BKCAL_OUT” enabled.
Enabling PID External Reset
• Utilized most often
in the primary loop
of a cascade
• Automatically
compensates for
poor secondary
loop response
Improving Process Recovery From
Saturated Conditions
• On recovery from a saturated condition, when
the ARW_HI_LIM and ARW_LO_LIM are set
inside the OUT limits, the reset will automatically
be increased by 16X until the OUT parameter
comes back within the the ARW limits or the
control parameter reaches setpoint.
Setting ARW limits
PV
SP
OUT
ARW_LO_LIM
OUT_LO_LIM
Eliminating Reset Action When
Control Error is Small
• By setting the
IDEADBAND
parameter to a
positive value,
then reset action
is held once
control error is
reduced below
this limit.
PID Non-Linear Gain Modifier
Knl
Knl=1
Knl=NL_MINMOD
e
NL_TBAND
NL_GAP
NL_HYST
Mode Determines the Source of
Output & Setpoint
Output Value
Select
Input
Parameters
(value + Status)
(Operator Entry
Setpoint Value
Select
Operator Entry
RCAS_IN
CAS_IN
Block Algorithm
(internal)
B
C
ROUT_IN
B
TRK_IN
C
BLOCK
AlGORITHM
A
SP
Value
D
Back-calculation
Parameters
RCAS_OUT
CAS_OUT
ROUT_OUT
D
Mode
A
Output
Parameter
OUT
Value
Mode Parameter Attributes
• Target is the requested mode
set by operator
• Permitted (configured) is the
selection available
• Actual is the achievable mode
given inputs status and target
mode
• Normal (configured) is the
designed mode
Rcas & Rout Modes
• If the RCAS_IN or ROUT_IN parameter is not
updated within a time defined by SHED_TIME
when in Rcas or Rout mode respectively, then
the actual mode of the associated block will
“Shed” based on the configured permitted
MODE and the SHED_OPT parameter.
Shed Options for Rcas & Rout Modes
AO Block
Analog Output Block
MODE
SP
OUT
CAS_IN
CAS
HI/LO
LIMIT
RATE
LIMIT
AUTO
(Output)
CONVERT
PV_SCALE
XD_SCALE
SIMULATE
BCKCAL_OUT
(Valve Position)
Readback
PV_SCALE
XD_SCALE
IO_OPTS
CONVERT
PV_SCALE
XD_SCALE
PV
I/O Options in the AO Block
• The Increase to
close option
should be set to
account for field
reversal so that
the SP value
always indicates
“implied” valve
position.
AO Setpoint Rate Limits
• The AO setpoint rate limits apply even when the block
is in CAS mode
(as well as Auto).
• This feature may be use to limit the maximum rate at
which a valve is change in automatic control.
• BKCAL_OUT status is set to limited if changes in OUT
are limited.
This prevents the PID from winding up under these
conditions.
Signal Characterization in Control Path
Using AO for Duty Cycle Control
• Percent time on
(ON_TIME) over
the duty cycle
period is
determined by AO
setpoint.
• Automatically
repeats at end of
duty cycle using
current value of
ON_TIME
• High resolution of
on-off time
ON
OFF
% time on
Duty Cycle Period
Duty Cycle Control – DO Setup
• Typical
application is
manipulation of
heater band input
for extruder
temperature
control
• Percise
adjustment to ½
of 60 hz since
timing is done by
DO card
hardware
• Percent time on is
determined by AO
output configured
to reference the
discrete channel
Duty Cycle Control – DO Setup
•
•
Period of duty cycle is
determined by
PULSE_PERIOD configured
for the DO channel
Period should be set to
match the period of
execution of the module that
contains the AO block that
references the DO channel
Duty Cycle – Configuring AO
• When you browse
to a DO channel
when configuring
the AO block, the
only selection is
ON_TIME
AO With Increase-Decrease Actuator
Continuous
Pulse Output
Channels
Proper Controller Tuning
• Is the fastest, quickest, and least expensive
improvement one can make in the basic control system
to decrease process variability.
• The detrimental effects of disturbances, interactions,
and control valve dead band are minimized by an
appropriate selection of tuning.
• There is always a tradeoff between performance and
robustness.
Uses State-of-the-Art Technology
– Process Identification Based
on Relay Self-Oscillation
Principle
– Applicable to a wide range of
processes
•
•
•
•
Slow
Fast
Self-regulating
Integrating
– Immune to Noise and Process
Load Disturbances
– Minimizes Tuning Time
Multiple Input - Single Output (MISO) Control
A multiple input - single output controller uses process inputs and outputs not
included in a single input- single output ( SISO) controller to improve control
response to disturbance and enforce operating constraints.
Information
needed to
Tune Controller
Feedback
Controller (SISO)
Controlled
Manipulated
Measured
Disturbance
(Other)
Process
Constraint
(Other)
Types of Process Variables
• Manipulated - process input which is adjusted to
maintain a controlled output at setpoint.
• Controlled - process output which is to be maintained at
a specific value; i.e., the setpoint
• Disturbance - measured process input which may also
affect the value of controlled outputs
• Constraint - process output which must be maintain
within an operating range by restricting the adjustment of
manipulated inputs.
First Order Plus Deadtime
Process response exhibits the combined characteristics of the lag and delay response.
O2
63.2% (O2 - O1)
O1
Output
Gain =
I2
O2 - O1
I2 - I1
Dead Time = T2 - T1
I1
Time Constant ( T ) = T3 - T2
Input
T1
T2
T3 Time
Integrating Response
Process output changes without bound when the process input is changed by a step.
O1
Output
I2
I1
Input
Time
Feedforward - MISO Controller
By immediately correcting for a measured load disturbance through adjustment
of the manipulated input, the control performance may be improved by
feedforward control.
Information
needed to
Commission
Feedforward
Feedback
L/L & DT
Dynamic
Controller
Compensation
+
+
L/L
DT
Manipulated
Measured
Disturbance
Process
Control
Feedforward Control
Measurable
Disturbance
Setup of L/L Block for Dynamic Compensation
Set the LEAD_TIME to Tm and the LAG_TIME to Td. The gain of the L/L
should be -(Load Dist Gain/Manip Gain).
Setup of Deadtime Block for Dynamic Compensation
The DEAD_TIME parameter should be set to a value of DT2 - DT1.
External Feedforward - Bias
• Feedforward
continues even
if PID is in Man
mode
• PID correction
is limited by its
output limits
External Feedforward - Bias(Cont)
• Act on IR should be
selected to allow the
bias (SP) to be
initialized on
transition from IMAN
to CAS actual mode.
External Feedforward - Ratio
• Feedforward
continued
even if PID is
in Man mode
• Ratio
correction by
PID is limited
by its output
limits
Override - MISO Controller
Automatic regulation of the process input to maintain one process output at target
without violating a constraint on another output is provided by override control.
Max Value
Override
Controller
<
Controller
Manipulated
Process
Information
Needed to
Tune Controllers
Over-ride Control
Control
Meas
Constraint
Meas
Unmeasured
Disturbance
Cascade Control
Secondary
Controller
Manipulated
Process 1
Disturbance
Disturbance
Primary
Controller
Process 2
Cascade Control
Unmeasured
Disturbance
Sec
Meas
Pri
Meas
Cascade –Dynamic Reset
• Select FRSIPID_OPTS for Dynamic Reset Limit
in the primary of the cascade to automatically
compensates for poor response of the
secondary loop.
• The CONTROL_OPTS in the secondary must be
set for Use PV for BKCAL_OUT for Dynamic
Reset Limit to provide benefit.
Split Range Control
Sec
Meas
Unmeasured
Disturbance
Split Range Control
AI
PID
TT104
TIC104
SPLT
FY104
AO
IP104A
AO
IP104B
TIC
104
FY
104
IP
104A
HEATER
IP
104B
COOLER
TT
104
Split Range Output (FY104)
100
Valve
Position
(% of Span)
Cooling (IP104B)
Heating (IP104A)
0
0
TIC104 Output (% of Span)
100
Splitter Block
OUT_1
100
OUT_2
OUT_ARRAY
0 100 0 100
IN_ARRAY
0 100 0 100
0
0
100
OUT_ARRAY
100 0 0 100
IN_ARRAY
0 40 35 100
100
0
100
0
100
LOCK_VAL “holds ”
LOCK_VAL “is zero ”
0
0
100
SP
OUT_ARRAY
0 100 0 100
IN_ARRAY
0 40 35 100
Splitter Block
• Total Cv and gain of 0-50, 50-100% split ranged
valves.
3-9, 9-15 Split Range
Flow Coefficient
16000
14000
12000
10000
8000
6000
4000
2000
0
0
10
20
30
40
50
60
70
80
90
80
90
100
Controller Output
3-9, 9-15
Split Range
5
Flow Gain
4
3
2
1
0
0
10
20
30
40
50
60
Controller Output
70
100
Splitter Block
• Total Cv and gain of 0-50, 25-100% split ranged
valves.
Flow Coefficient
3 - 9, 6-15 Split Range
16000
14000
12000
10000
8000
6000
4000
2000
0
0
10
20
30
40
50
60
70
80
90
80
90
100
Controller output
Flow Gain
3-9, 6-15
Split Range
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
10
20
30
40
50
60
Controller Output
70
100
Ratio Control
• IN_1 may be a
flow
measurement
(wild flow) or
setpoint of
another loop
• Ratio based on
SP will not reflect
deviations in flow
from setpoint
(resulting in
incorrect ratio)
Addressing Lag Dominated processes –
Fuzzy Logic Control
• The greatest advantage of fuzzy logic control is evident
when it is applied to processes with insignificant dead
time in order to accelerate the speed of control while
retaining a high-quality level of control.
• Where PID controller does not meet expectations, the
fuzzy logic control may be considered as an alternative.
Fuzzy Logic Control vs PID
Setpoint Change
Load Disturbance
FLC
FLC
PID
FLC
PID
FLC
PID
PID
Notice at both SP change and at load disturbance the FLC output change is more dramatic
than PID. Resulting in faster return to SP. Also notice as PV approaches SP, the FLC
exhibits less overshoot.
DeltaV Fuzzy Implemented As A Function Block
Pre-defined Fuzzy
Logic Control
Addressing Difficult Dynamics and Process
Interaction - Model Predictive Control
• Fully integrates DeltaV
Historian and off-line
process identification
– setup is trivial
• Model can be easily
updated.
• MPC with DeltaV is
easy
MPC -Addressing Difficult Dynamics
MPC
Temperature
Process
(1X1)
MPC -Addressing Difficult Dynamics
MPC -Addressing Difficult Dynamics
Automated process testing to
identify the process model
Step Response Model
Verification of identified model
Testing of control using simulated environment
Operator interface to MPC
Future Values of control
MPC Replacement for PID with Feedforward
Process
(2X1)
MPC
Measured
Disturbance
MPC Replacement for PID with Feedforward
MPC Replacement for PID Overrride
Process
(1X2)
MPC
Constraint
Unmeasured
Disturbance
MPC Replacement for PID Overrride
Addressing Process Interaction
MPC
Process
(2X2)
Addressing Process Interaction
Addressing Process Interaction
Layering MPC on Existing Strategy
MPC
RCAS_OUT
MPC
AI
PID
RCAS_IN
AO
Unmeasured
Disturbance
Process
(1X1)
Exposing RCAS_IN & RCAS_OUT
Right click on the
control or AO
block to expose
the RCAS_IN as
an Input
parameter and
RCAS_OUT as an
Output parameter.
Layering MPC on an Existing Strategy
Example DeltaV Predict Installations
•
•
•
•
•
•
Control Application
Industry
Evaporator
Chemical
Lime Kiln
Pulp&Paper
Pipeline Gas Blending Power
Bleach Plant
Pulp&Paper
pH Control
Pulp&Paper
Distillation Column
Pharmaceutical
Location
South Africa
Canada
Florida
Canada
US
Puerto Rico
Summary
• The function blocks in DeltaV may be used to address
plant requirement.
• The DeltaV MPC and Fuzzy Function blocks may be
used to better address difficult dynamic and process
interaction
• Your feedback on this presentation and input on future
control topic you would like to see presented at Emerson
Exchange are always appreciated.
terry.blevins@EmersonProcess.com
james.beall@EmersonProcess.com
Learning More About DeltaV Advanced Control
• Book was inspired by DeltaV
Advanced Control Products. This
book was introduced at ISA2002
may also be ordered through ISA,
Amazon.com or at
EasyDeltaV.com/Bookstore
• The application sections include
guided tours based on DeltaV
Advanced Control Products
• CD provides an overview video for
each section and examples.
Copies of the displays, modules,
and HYSYS Cases are included on
the CD.
DeltaV Predict and other DeltaV Advanced Control Products
Overview - Courses 7201, 7202, & 7203
These courses, beginning with the 7201, overview all of the major DeltaV
advanced control tools. Courses 7202, & 7203 each drill deeper into a specific
advanced control product and its application.
•
• DeltaV advanced controls are unique in the process control industry, in that
users do not need detailed knowledge of the underlying mathematical
principles to successfully apply the DeltaV advanced controls technology.
Course # 7201
DeltaV Advanced Controls
Overview
Course # 7202
DeltaV PredictPro
Implementation
Course # 7203
DeltaV Neural
Implementation
Emerson Exchange – Advanced Control Presentations
Effectively Addressing Control Applications
Monday 1:00-2:45, 3:00-4:45, Dallas 5-6-7
Addressing Multi-variable Process Control Applications
Tuesday 8-10, Grapevine D
Lime Kiln Optimization Using Supervisory APC – “Model
Predictive Control” & DeltaV
Wednesday 3:25-4:10, Texas 4
Adaptive Control – Better Control with No Tuning!,
Thursday 1:40 – 2:25 Grapevine A
DeltaV and Model Predictive Control - A primer on DeltaV
Predict and PredictPro
Thursday 8:40 – 9:25 Grapevine A
Field Experience in Property Estimation,
Friday, 8:00 - 9:40am & 9:55 - 11:35am Texas 5
Utilizing adaptive Control
Friday, 8:00 - 9:40am & 9:55 - 11:35am Texas 4
Download